1 / 22

Capturing, Modeling, Rendering 3D Structures

Capturing, Modeling, Rendering 3D Structures. Omnidirectional Cameras. Cameras. Traditional camera Pinhole Camera Model Problems: aberrations, distortions Tradeoff between aperture, shutter speed, focus, dynamic range Calibration Fit an assumed camera model to an actual camera

tillie
Télécharger la présentation

Capturing, Modeling, Rendering 3D Structures

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Capturing, Modeling, Rendering 3D Structures Omnidirectional Cameras

  2. Cameras • Traditional camera • Pinhole Camera Model • Problems: aberrations, distortions • Tradeoff between aperture, shutter speed, focus, dynamic range • Calibration • Fit an assumed camera model to an actual camera • Omnidirectional cameras • Single camera, multiple cameras, etc • Localization and pose estimation • Where is the camera relative to the object or environment

  3. A little bit of history… • Omnidirectional cameras are also called panoramic cameras • “Panorama” comes from the Greek phrase “all sight” • Originally used for artistic purposes • Robert Barker obtained a patent for the idea of a panorama in 1794 • “A Painting without Equal” • In 1800s, panorama became a common European word

  4. A little bit of Biology… • Some animals are capable of panoramic vision • e.g., certain insects, crustaceans (e.g., lobster) • Diurnal Insect Vision • Nocturnal Insect Vision • Crustacean Vision

  5. Taxonomy of Omnidirectional Camera Designs • Single center-of-projection • Like a traditional camera, light rays meet at a single “focal point” • Multiple center-of-projection • Camera does not have a single focal point • Sampled surfaces can be missing or duplicated in full image • Mathematical (re)projections are more complicated • Single Camera/Image • One “view” is acquired per image • Multiple Camera/Image • A single “view” composed by compositing several images

  6. Example Omnidirectional Camera Designs • Rotating camera design • Fish-eye lens design • Multiple camera planar mirror design • Single camera curved mirror design

  7. Rotating Camera Design • Place a camera on a tripod and spin it around snapping pictures every so often • Pros • Simple • Cons • Multiple centers-of-projection • Multiple (overlapping images) to composite • Vertical “jitter” • Slow acquisition process

  8. Rotating Camera Design original warped “stitched”

  9. Fish-Eye Lens Design • Use a wide field-of-view lens (~180 degrees) placed in front of a traditional camera • Pros: • Also relatively simple for users (making the lens can be troublesome for designers) • Cons: • Very severe image distortion • Low resolution around perimeter of field-of-view • Almost a single center-of-projection

  10. Multiple Camera Planar Mirror Design • Catadioptric = reflective (mirror) + refractive (lens) • http://www.fullview.com [Nalwa96]

  11. Single Camera Curved Mirror Design • Theoretical solutions to a single center-of-projection panoramic camera use mirrors that are subsets of swept conic sections • Cones • Spheres • Ellipsoids • Hyperboloids • Paraboloids

  12. Single Camera Curved Mirror Design • Theoretical solutions to a single center-of-projection panoramic camera use mirrors that are subsets of swept conic sections • Cones • Spheres • Ellipsoids • Hyperboloids • Paraboloids

  13. Conical Mirror

  14. Spherical Mirror

  15. Ellipsoidal Mirror

  16. Hyperboloidal Mirror

  17. Paraboloidal Mirror

  18. Catadioptric Paraboloidal Camera Motorized cart with camera, computer, battery, radio remote control [Aliaga01,02] Design by [Nayar97]

  19. Catadioptric Paraboloidal Camera z x focal point reference plane m mirror p image plane i Theoretical camera model

  20. Catadioptric Paraboloidal Camera reference plane m mirror ^ n p H lens image plane i Practical camera model [Aliaga01]

  21. - - ˆ ˆ n i m n p m × × = - - ˆ ˆ n i m n p m Catadioptric Paraboloidal Camera Calibration • Assuming incident equals reflected angle: • And given a 3D point p, mirror radius r, convergence distance H, we group and rewrite in terms of mr: mr5-prmr4+2r2mr3+(2prrH-2r2pr)mr2+ (r4-4r2pzH)mr-(r4pr+2r3Hpr) = 0

  22. Omnidirectional Vision Home Page • http://www.cis.upenn.edu/~kostas/omni.html

More Related